Overclocking CPU Cache?

Status
Not open for further replies.
It seems to benefit from memory related benchmarks and possibly software that heavily relies on memory bandwidth, though the actual speed increase is probably slim. In games, it makes pretty much zero difference. I can't say for sure it's "necessary", but I've read ideally you'd want your cache speed to be within 500mhz or so of your CPU clockspeed. I've read voltage should be ok under 1.3v
 

yumri

Distinguished
Sep 5, 2010
703
0
19,160
overclocking CPU cache espcially L1 cache is already going to do almost nothing so why not go for the best that you can tweak it too?
But yes i do agree that if 4.2 is stable but 4.201 and higher is not then stay at 4.2 as you want it to be stable more than anything else.
 

yumri

Distinguished
Sep 5, 2010
703
0
19,160
I kinda do not actually ussualy overclock on a work machine just on raspberry pis and the super cheap AMD based servers that i have built to which i can do it for that long to make sure that the overclock or underclock which ever is needed works properly without to much heat being generated by it.
 


Yeah du.

I still have a few questions. Does your system HAVE to be Stable with prime95 on the heaviest setting?? Because I get BSODs if I put prime95 on the most extreme stress test, but if I do a lighter stress test it doesn't bsod.

And how would I go overclocking the cache? is it the same as overclocking?

Sorry for all the questions, I can't find a lot on cache overlooking on Google.
 
Ok got it, and speaking of prime95 I did some research and found out that prime95 is a bad tester for haswell. So I just switched to OCCT and am 45min into a stress test on medium data set (what's the difference between medium and heavy?).

Ill get into uncore overclocking tomorrow, so should I start at a baseline of 1.2v on the cache at 42 and work my way down?

For now I have a stable 4.5Ghz at 1.23vcore on the CPU.
 
Just want to make sure one more time (just to be safe lol). It WILL BSOD if the cache voltage is insufficient right?

Just getting some weird behavior. When i have the cache voltage at 1.205 at 42 windows move around just fine while running the stress tester simultaneously, while at 1.180 at 42 i get 1-2 sec lags when moving a window with stress tester on too.

Nevermind, it's only happening on my CoreTemp window. My CPU-Z window and Thunderbird window move just fine.
 
Status
Not open for further replies.